Microprocessor Applications

Prof M.P.Gough m.p.gough@sussex.ac.uk

Syllabus :
Microprocessor architecture, Organisation & operation of microcomputer systems. Hardware and software interaction. Programme and data storage. Parallel interfacing and programmable ICs. Serial interfacing, standards and protocols. Analogue interfacing. Interrupts and DMA. Microcontrollers and small embedded systems. The CPU, memory and the operating system.

Teaching Methods: 2 lectures / week

Exercises/examples reviewed in workshops weeks 3,5,7,9

Research for handed in assignment (Week 10)

Assessment: Written Assignment Unseen Examination 20% Week 10 (March) 80% June

Reading List
Alan Clements. 2000. The Principles of Computer Hardware, Oxford, 3rd edition. (A number are available for loan from the Engineering & Design Department Office) For assessment exercise: Various manufacturer's microprocessor and microcontroller datasheets and user documentation downloadable from the internet. For lecture notes I have used the above Clements + others below Note: these are not recommended for buying The 68000 Microprocessor Family, M.A.Miller,1992 MacMillan Digital Fundamentals, Floyd, 2006, Pearson International Computer Engineering Hardware Design, M.Manno, Prentice Hall Microcomputer Interfacing, H.Stone, ddison Wesley + various datasheets from web e.g. 68HC000 =CMOS 68000 version

Course comprised of 8 topics:

1. Review Architecture & Programming of Microcomputer Systems 2. Programme and Data Storage 3. Parallel Input & Output Peripheral Devices 4. Interrupts 5. Serial Input and Output 6. Analogue I/O 7. Microcontrollers for Small Embedded Systems 8. CPU, Memory, and the Operating System

Typically Microprocessor within an Embedded System: Where Hardware meets Software

Principal function(s) controlled by microprocessor embedded within it Computer (microprocessor or microcontroller) hidden from view Purpose designed for particular application (PC is really a general purpose computing machine rather than an embedded system) Embedded computer takes input variables from controlled system Computes output variables to control system Sometimes autonomous, or sometimes interaction with user or sometimes interaction with other systems

The Microprocessor System Overview

In this course the 68000 or 68HC000 processor is used to demonstrate aspects of the device hardware interface and software device access

Part 1. Review Architecture & Programming of Microcomputer Systems:

- CPU architecture - 68000 example - Programming model and instructions (reminder of 1st year) - Microprocessor and the system bus - Connection to memory & I/O devices - Microcomputer organisation, signals and timing - System architectures

68000 Example. (or 68HC000)

Internal Software Program Model:

As a reminder of last years microprocessor programming Quick review of instructions follows.

Reminder: Move Instructions

A)MOVE General form: MOVE.<data size> <source effective address>,<destination effective address> Data size: B=Byte(8bits); W=Word(16bits); L=Long Word(32bits) Some examples of types of addressingRegister Direct : Address Register Direct: MOVE.W D2,D3 moves lower 16bits D2 D3

MOVEA.W D3,A0 moves lower 16bits with sign extension

Address Register Indirect: MOVE.L (A1),D0 32bits from memory pointed to by A1-> D0 Address Register Indirect with Displacement: MOVE.<size> displacement16(An),Dn MOVE.W $4(A0),D2 D2 memory at location given by (contents of A0 + 4) Absolute: Immediate: MOVE.L $C02E,D5 loads D5 with 32bit data word from location $FFC02E MOVE.L #$30,D2 loads D2 with immediate data (fills with leading zeros)

Address Register Indirect with Predecrement/Postincrement: e.g. MOVE.B (A3),D3, MOVE.L (A0)+,D4, MOVE.W D4,(A2)+

Reminder: Arithmetic & Logic Instructions

B) Arithmetic Instructions
ADD.<data size> <ea>,Dn Dn=Dn+<ea> and similarly for SUB (subtraction) MULU <ea>,Dn Dn <ea>lower word x Dn lower word similarly for DIVU (division)

C) Logical Instructions
ASL, ASR, LSR, AND, OR, NOT, Arithmetic shift left (lsb 0) Arithmetic shift right (old msb msb) Logical shift right (0 msb) Logical AND Logical OR all bits complemented 0 1

Reminder: Programme Control Instructions

D) Program Control / Branch BRA <relative address or label> unconditional jump in programme JMP <ea> unconditional jump to location specified by effective address Bcc <relative address or label> conditional jump where cc is flag condition. e.g BCC=branch if carry clear, BCS=branch if carry set Bcc often used after CMP compare two data values NOP no operation, time waster E) Use of Stack / Subroutines BSR, JSR unconditional branch/jump to subroutine (next programme address stack) RTS Return from subroutine previously saved on stack) ..plus many other instructions. (changes programme counter to value

Back to 68000 Programmers Model

Programme instructions intensively use the 8 data registers and 7 address registers in the CPU as intermediate data products or temporary variables in the course of processing data to / from the external world via external devices.

So where are these located relative to the typical system hardware? .

COMPUTER BLOCK DIAGRAM

Each device connects to: a) Data bus b) Address bus c) Control lines Control lines determine: i) signals timing for correct operation ii) device selection/activation iii) data flow direction. Only two devices allowed to communicate at any one time to avoid bus contention. Thus data move operations mostly one of these 4 types: i) Memory --> CPU or ii) CPU --> Memory or iii) I/O -->CPU or iv) CPU -->I/O.

[ Also a fifth type: Direct Memory Access I/O Memory but requires special DMA bus controller see later ]

External Hardware Connections

Summary of the 68000s 64 connection pins: Vcc Voltage source (e.g. 5Volts above Vss) Vss Ground Clock: system clock input Buses: D0 to D15 data bus lines - bidirectional A1 to A23 address bus lines, O/P Main Control lines: AS: Address strobe- valid address on A1-A23, O/P R/W: direction of D0-D15 bus,1=read,0=write, O/P UDS,LDS: upper/lower data strobe [A0], O/P [effectively A0:maps 8bit wide memories to 16 bits] DTACK: Data Transfer Acknowledge, I/P slower external devices can cause CPU to wait RESET: resets CPU programme counter I/P HALT: halts operation(I/P) or indicates failure(O/P) IPL0-2: Interrupt request lines I/P Others: BR,BG,BGACK: for external DMA control of bus FC0-FC2:monitor: programme, data, interrupt ,O/P E,VMA,VPA,BERR extra signals for interfacing

Programme Memory Read Cycle (Main signals in bold)

Main aspects: FC0-FC2=010 indicates program opcode fetch (alternative 001 for data) Valid address A1-A23, UDS,LDS=00 means16bit read (10=d0-7, 01=d8-15 only) Address Strobe, AS allows address bus to be decoded for memory chip select R/~W stays high throughout as this is a read operation External device/address decode asserts ~DTACK as data placed on bus If memory device is slow ~DTACK assertion can be delayed to provide wait states D0-D15 Data bus receives valid data from addressed memory before AS returns. 68000 uses a 2-word prefetch, absorbing program fetch cycles within execution cycles

Memory Write Cycle

(main signals in bold)

Main aspects: FC0-FC2=001 data memory.

Valid address A1-A23, UDS,LDS=00 means16bit write (10=d0-7, 01=d8-15 only). Address Strobe, AS allows address bus to be decoded for memory chip select. R/~W goes low to indicate this is a write operation,writing D0-D15 to memory. External device/address decode asserts ~DTACK as device reads data from bus. If memory device slow ~DTACK assertion can be delayed to provide wait states.

System Design
1) Before Designing system decide on requirements: Amount of programme memory (ROM) Amount of read/write data memory (RAM) Number & type of I/O ports Other system and peripheral components as needed 2) Software must be considered. 3) Then individual component types chosen, considering their characteristics (timing, voltage levels,etc) & requirements 4) Circuit wiring, board design & board layout completed

Part 2. Programme & Data Storage

- Types of memory device - Connecting memory to the processor - Memory device address decoding

Types of Memory
Random Access Memory, RAM (data volatile- lost on power off)
RAM used for data, can be written to & read from Static RAM each bit stored in simple circuit of a few transistors, e.g. flip-flop Dynamic RAM- each bit stored as charge on a single transistor gate but needs refresh circuitry as gate is a leaky capacitor and data lost otherwise SRAM faster, takes more power, less dense expensive, but easy to use DRAM simpler, lower power, cheaper, requires extra refresh control, more complex to use.

Read Only Memory, ROM (data non-volatile, remains after power cycling)
ROM data remains after power off. Mask programmed custom written at manufacture, e.g. PC boot up programme PROMS semi-custom- written only once to chip by specialist equipment/co data 0/1 stored as fuses blown/unblown or as OTP (see below) EPROM user programmed by EPROM programmer. Data stored as charge on high impedance gates- can be erased by ultra-violet light through window in chip & reprogrammed. One time programmable, OTP, = version of EPROM chip without window EEPROM- similar to EPROM but erased electrically without being removed from circuit. Erased in blocks of memory in system programmable Flash memory, similar but simpler very dense memory (silicon hard disc) FRAM access as fast as RAM but data non-volatile

Address Decoding
a) General address decoding Chip selected by specific combination of higher address line values.

b) Linear Address Decoding Each chip select uses a dedicated address line- simple for small systems but wasteful and can lead to bus contention (>1 device selected at once!! e.g. A11 & A12 must not both =1 ) c) Full Address Decoding Logic used to provide a maximum number of chip selects from address lines. E.g. two address lines A11 & A12 have four possibilities (00,01,10,11) each combination decoded for a chip select.

Full Address Decoding: Decoder Chip 74138

When chip is not enabled: all 8 outputs high independent of A inputs When chip enabled (~E1,~E2,E3=001) only one output goes low, rest high Inputs A1,A2,A3 select which of 8 outputs goes low

Programme memory : ROM/EPROM

M6836 16k x 8 Byte wide Data: DQ0-DQ7 Address: A0-A13 ~G is Read ~E chip select

Two examples:

Intel 27210 64k x 16 16bit word size Data: O0-O15 Address: A0-A15 ~CE is chip select ~OE is enable ouput (read data from ROM) ~PGM for programming data into ROM

16 bit wide ROM:

D0-D15 ROM uP All uP A1-A16 ROM A0-A15 64k x 16 ROM ~CE from 138 decoder when A17,A18,A19=000 Other combinations for other devices As ROM all accesses are read so ~OE=~CE ~DTACK low while ROM selected
A19 A18 A17 A16 A15 A14 A13 A12 A11 A10 A9 A8 A7 A6 A5 A4 A3 A2 A1 0 0 0 0 0 1 0 1 0 x x x x x x x x x x x x x x x x ROM 00000H-0FFFFH 10000H .. Next Device

Another Device20000H

Using two 8 bit wide ROMs for 16bits data bus

A0= UDS/LDS A1-A14 micro both ROMs A0-A13 ROM1 = D8-D15 ROM2 = D0-D7 Address decoder selects ROMs for A16-A18=0 ROM: 0000H-03FFFH other 138 outputs used for other devices, RAM etc

Connecting RAM
Addition of two 32K x 8 RAM to previous slide ( two of ROM of last slide not shown for clarity ) Again pair for 16bits wide ROM A16-A18=000 RAM A16-A18=001 Now R/~W needed ~DTACK as long as either ROM or RAM accessed.

Part 3. Parallel I/O and peripheral devices:

- Buffers and latches - Example input and output devices - Programmable I/O devices - Counter-timers

Buffers for digital input port:

Buffers enable to pass at specific times. Single buffer: pin 1 when low passes data from pin 2 to pin3, otherwise high impedance on pin3. Combining 8 buffers in parallel provides a means for digital input. When common output control low, data from lines I0 to I7 is passed onto data bus for microprocessor to use (INPUT). ~Output control = logical OR of ~PortCS, ~Bus Read

Latches for digital output port:

D-type latches used to sample & hold data - latch data. Data on the bus, e.g.from the microprocessor, sent to the output port. Address decoding for port enables latch, latching data at end of pulse. Port outputs new data until next time port addressed (OUTPUT). ~LE = logical OR of ~Port CS, ~Bus Write)

Octal latch can be 8bit Port

Peripheral Interface Adaptor(PIA)

Example of MC6821 (simple 8 bit port)
Two Ports: A & B Each port has 3 registers: 1)Peripheral Data Register buffers actual port data 2)Data Direction Register each bit 0 for I/P, 1 for O/P 3)Control Register sets condition for data flow, interrupt, & handshake with external devices

Connecting the PIA

Connecting to previous system of ROM & RAM, using decoder set for A16-A18=110 and A0=0 for port. Lower address bits A1, A2 select PIA registers for I/O data flow Control. Interrupt requests connected to 68000 interrupts so that each Port A & B can be processed by Separate interrupt routines.

PIO Handshake with external world

Input- keyboard example:

Keyboard Key pressed CA1 high Keyboard tells PIA that data ready to be read. CA1 used to strobe data into PIA data register. PIA acknowledges keyboard by CA2 high and at same time tells uP by setting IRQA low Microprocessor services interrupt request and reads PIA data register. Act of reading resets IRQA high and resets CA2 low telling keyboard that it is ready for more data.

Keyboard Encoder Example:

Rows scanned with travelling 0 on output port until keypress causes input <11111111 Then key identified by combination of known position of 0 on O/P port and measured position of 0 on I/P port.

Code for keyboard scanner/reader

ORG $002000 Xlines EQU $008000 Ylines EQU $008002 MOVE.B #%01111111,D0 MOVE.B #-1,D1 XLOOP ROL.B #1,D0 ADD.B #1,D1 AND.B #%00000111,D1 MOVE.B D0,Xlines MOVE.B Ylines,D2 CMP.B #%11111111,D2 BEQ XLOOP CLR.B D3 YLOOP CMP.B #%11111110,D2 BEQ JOIN ROR.B #1,D2 ADD.B #1,D3 BRA YLOOP JOIN LSL.B #3,D3 OR.B D3,D1 RTS ;Output port for rows ;Input port for columns ;Initial X value with 0 ;Preset X counter = -1 ;Rotate position of 0 ;Increment X counter ;X counter is modulo 8 (values 0-7 only) ;Output X value to keyboard ;Read Y value from Keyboard to D2 ;Any 0 in Y Any key pressed? ;Repeat until key pressed ;Preset Y counter = 0 ;test for a 0 in lsbit of D2 ;Exit to concatenate X & Y values ;Rotate D2 one place right ;update Y counter ;test next bit position for 0 ;Shift Y counter 3 places to the left ;Add in X counter ;Return value in bits: 00yyyxxx ( 0-6310 )

Counter-Timer Chips
This example three separate counter-timers Each has clock input, a gate input, and an output: a)Clock can be supplied from the microprocessor clock, or by an external system. b) Gate is a signal that enables/disables counting c) The output is changed when the counter reaches a preset value, counted down 0. Uses: Output can be used as interrupt to uP Enables accurate time delays to be generated under software control Multi-Mode configured by software Used for delay instead of software timing loopsfrees up uP to do other tasks Can be used to count external events Watchdog timer- unless software reloads counter before an initial long count value reaches zero resets system. Checks against endless loop type software hangups - ensures continued operation of essential systems.

Examples of two of the many Modes provided by a Counter/Timer Chip:

Other modes: Mode 1- Programmable One-Shot, Mode 4- Software Triggered Strobe,

Mode 2- Rate generator Mode 5- Hardware Triggered Strobe

Part 4. Interrupts: - Need for interrupts

- Principles of interrupt-driven I/O - Interrupt programming techniques - Interrupt Priority & Interrupt Vectors

Output to Port with fixed software delay (without interrupts)

Port Count Deloop : MOVE LEA LEA : LOOP1 MOVE.B MOVE.B JSR SUB BNE : Delay Loop2 MOVE SUB BNE RTS ORG DS.B #Count,D1 ;set up loop counter Table,A0 ;A0 points to table in memory Port,A1 ;A1 points to Port (A0)+,D0 D0,(A1) Delay #1,D1 LOOP1 ;D0memory([A0]) ; [A0][A0]+1 ;Output data ;decrement loop count ;repeat for all 128 data EQU EQU EQU ORG $01800 Location of Port 128 Size of block to output 64 wait loop $000400 Program origin

PSEUDO-PROGRAMME:

FOR i=1 to 128 move data from table to port wait a fixed time END FOR

Disadvantages: Need delay time between

#deloop, D2 ;set up delay loop time #1,D2 ;decrement loop time Loop2 ;wait for loop time ;return from subroutine $002000 128

outputs to be sufficient for external devices. No handshake used

Table

;128bytes reserved for data table

Microprocessor tied up by programme while waiting

Output to Port with polling (without interrupts)

Portdata Portstat Count EQU EQU EQU ORG : MOVE LEA LEA LEA : : LOOP WAIT #Count, D1 ;set up loop counter Table,A0 ;A0 points to table in memory Portdat,A1 ;A1 points to Port data Portstat,A2 ;A2 points to Port status $08000 $08002 128 Location of Port data Location of Ports status byte Size of block to input

PSEUDO-PROGRAMME:

$000400 Program origin

FOR i=1 to 128 get data from table wait until port ready output data END FOR

MOVE.B MOVE.B AND.B BEQ MOVE.B SUB BNE

(A0)+,D0 (A2),D2 #1,D2 WAIT D0,(A1) #1,D1 LOOP1

;D0memory([A0]) ; [A0][A0]+1 ;Read status ;mask off all but ready bit ;wait for port ready ;Output data to peripheral ;decrement loop count ;repeat for all 128 data

Disadvantages: Limited handshake Microprocessor tied up waiting for peripheral to be ready

: : ORG DS.B $002000 128

Table

;128bytes reserved for data table

Need Interrupts...

Interrupt Driven I/O

Each interrupt vector to subroutine which: Gets pointer for next entry, Reads a byte, Outputs to port, Moves pointer to next entry, Saves pointer in memory, Returns from interrupt

OUTPUT : INT Y

EQU ORG

$008000 $000400

Location of O/P Port Start of programme Save environment general for subroutines Point A0 to buffer Read a byte from buffer Send to O/P port Save updated pointer Restore Environment Return from interrupt

MOVEM.L D0-D7/A0-A6,-(A7) MOVEA.L POINTER,A0 MOVE.B (A0)+,D0 MOVE.B D0,Output MOVE.L A0,POINTER MOVEM.L (A7)+,Do-D7/A0-A6 RTE $002000 1024 BUFFER

: ORG BUFFER DS.B POINTER DC.L Data Origin Reserve 1024 bytes Reserve long word

In previous example (of last 2 slides): to obtain regular slow timed outputs- interrupt could be caused by a software pre-programmed Timer/counter chip output connected to a processor interrupt line.

Interrupt: Priority & Vectors

Interrupt Priority. Example of 68000 has 7 levels of interrupt priority: 3 input pins IPL0-IPL2 can have values 0-7 (values negative logic) 0=no interrupt, 1=lowest priority interrupt 7=highest priority level interrupt. All interrupts at level 3bit mask in 68000 status word are serviced Level 7 is thus a non-maskable interrupt - always serviced Software can control when to service Interrupts < level 7 e.g. dont interrupt time critical processes Interrupt Address Vectors (Interrupt programme control sequence) Peripheral provides interrupt signal to Processor Processor acknowledges to peripheral that it will accept interrupt Peripheral provides interrupt vector to processor Processor uses vector to look up location of interrupt handler routine

Multi Peripheral Interrupt + Acknowledge

Each Peripheral:
provides interrupt receives acknowledge

Priority Encoder:
converts IRQ1-IRQ7 to three bits IPL0-IPL2

IACK Decoder: decodes CPU response

function code = IACK + address = level & generates IACK1-6

Peripheral with IACK:

Provides interrupt vector e.g. 40H

CPU:
gets interrupt vector from memory pointed to by peripheral vector x 4 e.g =100H

Part 5. Serial I/O:

- Asynchronous and synchronous transmission
- UARTs

- Serial I/O under program control - Other standards

Serial Interfaces
Serial Transmission can be: i) Asynchronous ( e.g. traditional PC COM1 port ) ( e.g. USB Port )

ii) Synchronous

UART chip performs parallel-to-serial conversion on data sent from CPU and serial-to-parallel conversion on data received by CPU. Mechanism of shift register, shift out bits of data byte (or character) one at a time.

Bit-Serial Data

Non-Return to Zero (NRZ), quiescent level=1 Bit serial data framed by start and stop bits with optional parity bits for error checking. E.g. if 7bit character data(ASCII) then up to 11 bits required per character. Above example of transmitting character R, in ASCII is 52Hex (1010010b). Seven bit data. (8th MSBit discarded). Character rate = bit rate / bits per character Bit rate = baud rate

Data Link can be: 1) Simplex ( one way data transfer ) 2) Half Duplex ( two way data transfer, but only one at a time ) 3) (Full) Duplex ( two way data transfer simultaneously )

Typical example of serial communications

Extra signals needed for handshake with external serial devices: RTS: Request to send. Computer asks modem if it is ready for data operations CTS: Clear to send. In response to RTS modem tells computer data can be sent DCD: Data carrier detect. Modem tells computer that it receives carrier tone on the telephone line

UART for 68000 - Asynchronous Communications Interface Adapter)

Using the ACIA: Example Software

Configuring ACIA: ACIA EQU $800000 CR EQU 0 LEA ACIA,A0 MOVE.B #%00000011,CR(A0) MOVE.B #%10110101,CR(A0) ;ACIA address ;Control Register Offset ;A0 points to CR ;software reset (user looks up codes in datasheet) ;set baud rate, handshake, interrupt ( )

Receive a Character Subroutine: RDRF EQU 0 ;RX data ready bit 0 of SR SR EQU 0 ;Status register offset DR EQU 2 ;Data register offset LEA ACIA,A0 ;A0 points to ACIA POLL TST.B #RDRF,SR(A0) ;Read RX status bit BEQ POLL ;repeat until char received MOVE.B DR(A0),D0 ;get input from ACIA to D0 RTS Transmit a Character Subroutine: TDRE EQU 1 LEA ACIA,A0 TPOLL BTST.B #TDRE,SR(A0) BEQ TPOLL MOVE.B D0,DR(A0) RTS ;Transmitter data register empty bit ;A0 points to ACIA ;TX register empty? ;Repeat until ready to transmit ;Move byte from D0 to ACIA

Types Of Serial Interface:

RS232C: (Bipolar) Logic 1 has value < -3V (typically -12V) Logic 0 has value > +3V (typically +5V). Low impedance 50

RS423:

RS422:

Low impedance differential twisted pair eliminates ground loop pickup,etc

RS485:

Universal Serial Bus, USB

fast data + can power small devices from bus e.g. flash memory

Some other Standards:

Firewire ( IEEE 1394 )
External serial bus for fast data transfer 400Mb/s (developed by Apple). Up to 63 devices can be connected in daisy-chain arrangement. 6 wires: two twisted pair for data , power, ground

GPIB, General Purpose Interface Bus ( IEEE488 ).

Parallel Bus developed by Hewlett-Packard for test & measurement equipment 1MByte/s. 24 lines: 8 data lines, 8 ground returns/screening, 3 handshake lines, 5 bus-management lines.

SCSI, Small Computer System Interface (scuzzy).

Parallel Bus with various variations & connectors e.g. a) SCSI-1- 25 pin connector 8-bit data + handshake, upto 4Mbytes/s b) Wide Ultra SCSI-2, 16bit data at 80Mbytes/s

Part 6. Analogue I/O (or Digital meets the real World):

- Digital-to-Analogue (DAC) principles - Analogue-to-Digital (ADC) principles - Software Interfacing methods - Sampling and aliasing - Programming techniques - Introduction to digital filtering

Potential Divider Network DAC

Many resistors needed - 2n where n= number of bits, but all same value

D to A: Type= Binary-Weighted-Input DAC

a) Simple Explanation: Each bit if logic 1 connects resistor to the circuit. R values vary as 2:1 from bit to bit with MSB having the lowest value R low R passes highest current most effect on output voltage. Disadvantage: need to have many different, precise R values

b) Detailed Explanation: Amplifier -ve input is virtual ground since feedback resistor from Vout holds inputs at 0 volts. High impedance input amplifier takes zero input current, so all currents I0, I1, etc, must pass through feedback resistor, Rf. Total current in feedback resistor, If = b0 I0 + b1 I1 +b2 I2 + .. ( With each bit bn =0 or =1) So final analogue voltage output= Vout = If Rf

An R-2R Ladder type DAC

Advantage: only two different R values: R & 2R

Example: data value 1000 D3=1 5V via 2R to input held at 0V by feedback- all current flows through Rf(=2R) so Vout must be -5V. Lumped value, Req of other resistors not critical as no current passes through Req

R-2R ladder DAC- another example value = 0010

Thevenins Theorem- any circuit can be reduced to an equivalent voltage in series with an equivalent resistor. Applying theorem to the left of R8 we have Vth=1.25V & Rth=R. Again voltage across R7=0, then 1.25V through 2R to input will require Vout to be -1.25V to keep input at zero voltage (Virtual earth).

a) Square wave period T (without DAC): Loop1 Output 0 on port pin wait T/2 Output 1 on port pin wait T/2 branch to Loop1 repeat forever b) Sawtooth ramp period T (with 8 bit DAC) initialise D0=0 Loop2 output D0 to DAC increment D0 wait T/256 branch to Loop2 repeat forever c) Triangular wave period T (with 8 bit DAC) initialise D0=0 Loop3 output D0 to DAC increment D0 wait T/512 compare D0 to #255 branch to Loop3 if not equal Loop4 output D0 to DAC decrement D0 wait T/512 compare D0 to #0 branch to Loop4 if not equal branch to Loop3 repeat forever

Waveform Generator Pseudo-Programmes

256 steps up/down

DAC performance aspects:

Resolution
Improves with number of bits, n. % resolution = 100 * 1 / (2n 1)

Accuracy
Ideally = resolution but in practice less because of accuracy of resistors

Linearity
Linear error is deviation from ideal straight line Vout= constant x digital value

Settling Time
Time taken for analogue output value to reach a new value in response to a change in the digital input - depends on RC time constants , internal & external capacitance.

Analogue to Digital Conversion: Successive-Approximation

Example of 4bit ADC 1) Microprocessor sends SC (start conversion) to control 2) Control logic within ADC outputs a digital value to a DAC 3) The analogue input is compared with the DAC output 3) Control logic tries each bit in turn starting at MSB Decision tree: Only four decisions (red lines) 4) After 4 successive approximations sends end of conversion EOC signal to microprocessor 5) Microprocessor reads digital value

Flash ADC 3-bit example

Input analogue signal fed in parallel to many comparators which compare input against a voltage divider chain. Bits set from bit 0 to the voltage tap just below the input voltage value. An encoder then converts the signal to binary value. e.g. Vin = Vexample 7 .1 0000111 0112

Vexample

Dual Slope ADC

Start: assume counter is zero and output of integrator=0. 1) Switch connects +ve Vin to R assume Vin steady (sample/hold) 2) C charges linearly(const I=V/R) resulting in negative ramp at V0 3) When counter reaches a preset value (time T) counter reset, & control switches input to -Vref causing C to discharge linearly towards zero. 4) When Vo reaches zero comparator stops count. 5) Count value t1 or t2 depends on size of the input voltage. 6) Binary count t1/t2 read out

Main ADC Performance Aspects:

Conversion Time (application specific & the need to avoid aliasing) Conversion Accuracy (increases with number of bits) Electrical Power consumption may be limited in small systems (power increases with conversion rate)

Example software to access ADC

ADCstatus ADCdata Size Table EQU $8001 EQU $8000 EQU $80 EQU $4000 MOVEA.W A0,ADCstatus MOVEA.W A1,ADCdata MOVE D1,$Size MOVEA.W A2,table Loop Wait MOVE.B $01,(A0) MOVE D0,(A0) AND A,$01 BNZ Wait MOV (A1),D0 MOV D0,(A2)+ SUB $01, D1 BNZ Loop ADC status register address ADC data register address Number of values to read into table Address of destination table in memory A0 points to ADC status register A1 points to ADC data register D1 holds the number of values to read A2 points toTable of values read from ADC Start ADC Conversion- set SC bit Read ADC Status Mask off bits other than EOC bit Wait for End of Conversion EOC Read ADC value Store in table, increment table position Decrement Loop counter Repeat to complete data table

Sampling & Aliasing Error

a) For good reconstruction of signals the sampling frequency, fsam, should be > 2 fmax, where fmax is the maximum signal frequency or, fmax Nyquist frequency (= fsam/2 ). Example of sufficient sampling: figure on left. b) Absolute limit of 2 samples per wave cycle fsam = 2 fmax (figure lower left). c) Aliasing errors occur when fsam< 2 fmax as illustrated here in the figure below right. i.e. less than 2 samples per wave cycle. The reconstructed waveform is then a very different frequency from the original signal.

Simple Digital Processing Example: e.g. moving average filter

Analogue signal ADC Digital Processing Typical Filter Processing: DAC Processed Analogue signal

x(n)=sampled analogue waveform, an =weights (coefficients, or scaling factor), Z-1 =unit time delay = 1 sample period

Simple Digital Filtering

Moving Average FIR Filter: Specific ADC input, x(n) {processing} result, y(n) DAC

y(n) = (1/4){ x(n) + x(n-1) + x(n-2) + x(n-3)}

Use four registers D0,D1,D2,D3 to store signal samples x(n), x(n-1), x(n-2), x(n-3) LOOP Read new ADC value Store this new value in D0 Add D0 to D1 Divide by 2 (arith shift right) & store in D4 = (1/2){ x(n) + x(n-1)} Add D2 to D3 Divide by 2 & store in D5 = (1/2){ x(n-2) + x(n-3)} Add D4 to D5 Divide by 2 Output this value to DAC = (1/4){ x(n) + x(n-1) + x(n-2) + x(n-3)} Move D2 contents to D3 x(n-2) x(n-3) Move D1 contents to D2 x(n-1) x(n-2) Move D0 contents to D1 x(n) x(n-1) Repeat LOOP forever get new x(n) Note: No need to initialise D1-D3 as not important after 3 programme loops.

Part 7. Microcontrollers for small embedded systems:

- Configurations - Architectures - Features - Other aspects

Comparison of microcontroller with microprocessor

Computer Architectures:

Most microprocessors use von Neumann architecture as Harvard would need many more pins to access two external buses. However, more processing efficient Harvard Architecture with two buses easily implemented internally within a microcontroller.

Multi-Processor / Parallel Processing

Microcontroller Example. An industry standard : 8051

Dont worry too much about complicated schematic on left. Mainly to show here the very many features included within one chip. Main ones are highlighted: Ports0-3, RAM, ROM, ALU, Oscillator, serial port, timer/counter, etc

8051 chip Includes: central processor ROM & RAM 3 counter/timers 4 parallel ports 1 Serial port Requires only crystal for clock and Vcc. Ports can be used to expand ROM & RAM bus externally

8bit microcontrollers:

Some Microcontrollers

Very many types & manufacturers produce various versions with different facilities, e.g: a) Speed: reduce 1 clock / instruction b) Memory Data RAM upto 2kbyte Programme ROM upto128kbytes+ Types: EPROM, Flash, EEROM. c) Communications RS232 I2C, 1-wire CAN bus, Ethernet, etc d) Peripheral Drivers, LCD, etc

16bit microcontrollers:

Multi-Core and Parallel Processing

10.0

Recent progress in processor speeds curbed by power dissipation problems. Every transistor switch action has I*V. Very many transistors at any one time often doing nothing but still dissipate heat! Over-clocking say from 3GHz to 3.6GHz possible beyond manufacturers specifications running hot reduces component lifetime plan redundancy use 10year guarantee computer for only 4year Supercomputers- similar problems. Germanium Arsenide logic + exotic components + liquid cooling faster clocks. Supercomputer progress only through vastly parallel machines with many processors (groups of 1000s of PCs) - massively parallel hardware and applications Desktop PCs co-opted supercomputer model single die. Dual core , Quad core 2 to 4 CPU cores on a

Intel Core 2 Duo 64bit dual core

10.1

Typical 2 CPUs each with own L1 cache, share single L2 cache, that accesses single external bus to off-chip memory (Cache memory- see later) But typically only ~1/3 of presently written PC programs can be parallelised only 50% speedup as go from single dual core (not the expected 100% - Amdahls law) Single bus from L2 to off chip external RAM memory still a bus bottleneck Programs only fast as long as work from L1. But L1 size typically only 32-64kbytes! Small programs!, L2 typically 4Mbytes Multi-core (2 -4) use is presently optimised by operating system: multiple programs written for single core. Primarily for speeding up Multi-tasking, Multiple threads Moving towards Many-cores >>4, Many-cores really need to program specifically for parallel processors, need effective parallel languages, auto-parallelising compiler Active Research Area

CELL Broadband Engine multi-processor for Playstation3

10.2

Overview: 9 processor elements on a single chip: 1 x 64 bit PowerPC processor element (PPE) optimised for operating system/control 8 x Synergistic processor elements (SPE) optimised for compute intensive applications PPE access main storage via load/store to private register file. Operating system neutral SPE access main storage via DMA to local memory for data & programme See: http://www-01.ibm.com/chips/techlib/techlib.nsf/products/Cell_Broadband_Engine

10.3

Xera flop 1027 Yotta flop 1024 Zetta flop 1021 Exa flop 1018 Peta flop 1015 Tera flop 1012 Giga flop 109 Mega flop 106 Kilo flop 103

Flops History: Logarithmic plot

Today: 1) Top of the range PC Quad-core 30Gigaflops; 2) Roadrunner super computer (~ 130kcores = 13k Cell processors[9core]+ 7k AMD dual core)

1.7 Petaflop

Future/Now: Processors as CORES in FPGAs Field Programmable Gate Arrays

Gate arrays- a sea of uncommitted logic cells can be configured as complex system that includes several microprocessors. For example, a single Xilinx Virtex 4 family FPGA chip can include Two PowerPC 32bit RISC processors 192 DSP slices (multiply-accumulate units- for signal processing) 4 x 10/100/1000 Ethernet interfaces 142K Logic cells Block RAM etc

FPGAs can be re-configured in application to provide various functionalities. e.g. mobile phone, GPS receiver, etc. FPGA cores often operate at lower voltage than data buses, often mixed voltage buses. therefore there is a need to convert logic levels between buses..

Multi-Voltage Level Systems

Systems often utilise more than one voltage level logic to optimise the system by making use of the most appropriate chips. Various logic level standards: 5V, 3.3V, 2.5V, 1.8V, 1.5V, etc. Lower voltage faster + lower power dissipation Need bidirectional bus transceivers with voltage level shifting between different voltage buses. Example below where, say, bus A is 5V logic and bus B is 3.3V logic. Chip with 8 or 16 data lines, each connected as shown here. Buses can be isolated or joined by ~Enable line, while data left-right direction is set by Direction line.

Part 8.

Other System Aspects:

- Direct Memory Access - Cache memory - Operating systems & Multi-tasking - Connecting to sensors & actuators I2C, 1-wire, CAN - Programming, cross compiling, system debug

Direct Memory Access, DMA:

for

Fast I/O

Memory

Direct Memory Access, DMA, contd

DMA modes:

Generous!

Reasonable

Greedy!

CACHE MEMORY For Faster Programs

7.0

Memory Cache for faster programmes

Cache memory local fast memory used to hold pre-fetched operation codes Speed-up depends on (i) ratio of cache memory speed to main memory speed, (ii) how often op-code is already in cache (a hit), (iii) average number of machine/clock cycles per instruction Cache controller needs to (a) look ahead in programme to fetch instructions. (b) keep address tags of instructions to identify hits Note that the 68000 uses a standard simple 2-word pre-fetch, absorbing some program op-code fetch cycles within execution cycles as many clock cyles/instruction.

Operating System
Overall OS: Co-ordinates, optimises efficiency, schedules tasks (processes). Applications use resources provided by OS OS hides details of the hardware.

Task Scheduling: Each process is in one of three states: Runnable: available & waiting Running: running now Blocked: waiting for an essential resource to become available.

Multi-Tasking
OS safely switches contexts between processes. Scheduler saves current processs context (volatile portion) and invokes a new Process. Present state of each process must be saved at end of running it. Previous state of each process must be restored at the start of running it again. Use separate stack areas for each process to save register status.

Connecting to other systems

Previously we mentioned RS232, USB, Firewire, SCSI, etc as main standards for microprocessor / computer connection to peripherals. Main standards for microprocessor/controllers networking to sensors/actuators:

CAN, Controlled Area Network

Balanced 2-wire interface with differential line drivers / receivers (like RS485) Used in Automobile, Transport & Industry for up to 100m communications. e.g. Automotive Bus, Industrial Field Bus

I2C,
Fast 2-wire bus, up to 400kbits/s

1-wire,
Single wire used by master to communicate with slaves, also used to power slave devices. Economic in hardware resources. Ideal for short distances.

CANbus
Each byte transmitted as NonReturn to Zero, NRZ, asynchronous, with start & stop bits (like RS-232.). Balanced 2-wire interface with differential line drivers & receivers in parallel (like RS422/RS485).

Data sent in frame:

Start of Frame Arbitration Control: 11-29 bits determine priority of message, arbitrates between devices. Data: 0-8 bytes of data Cyclic Redundancy Check: 15 bit checksum Acknowledge: any CANbus device receiving acknowledges TX retransmits if none. End of Frame

I2C Bus
Each device on bus has unique address. Multi-master- more than 1 device can control bus. Arbitration between contending devices. Serial 8bit data. Two wire bus shared by all devices: SDA Serial Data line;SCL Serial Clock Line Example:

1-wire

(Maxim-Dallas)

Device families include: ADC, DAC, Analogue Switches, Memory, Temperature Sensors, etc.

Programming Microprocessors/Microcontrollers
Directly in Low Level Assembler Language. Slow, tedious, unforgiving, only practical for small systems Timing for critical programme loops and for interfaces can be set precisely from number of clocks/intruction. Memory use/allocation can be easily organised/kept within bounds. Better to understand what is actually happening in fine detail. Difficult to appreciate the whole design. Indirect via Cross Compiling from Higher Level Language, e.g. C Relatively quick, easy to implement in C, often necessary for large systems Difficult to ensure the precise timing of critical parts. Care must be taken in memory use and data variable type assignment. Easy to appreciate the whole and verify overall design functionality. In practice overall system often written in a higher language with some time critical sub-systems written directly in the relevant assembler language.

System Debug Tools

ROM Emulator.(software development in real system)
Replace programme ROM in its socket by lead to PC which provides code CPU and remainder of target embedded system as is. Allows code to be run from any given address up to user selectable breakpoints.

In-Circuit Emulator, ICE (real in-system verification)

Extract CPU from socket (or attach adaptor & tristate CPU) and replace by ICE system. ICE provides as for ROM emulator + more control: permits full view of system, signals and bus devices with full trace facilities.

Logic Analyser (real system hardware debug)

Multi-channel, multi-signal digital oscilloscope / monitor, specially for detailed analysis of system bus. Most useful for debugging specific problem events, e.g. system timing of communications between devices.

Simulator (virtual system)

Software simulation no use of actual hardware.

Course Summary

Problem Sheet 1: Address Decoding

A 68000 microprocessor with address lines A1-A23 is to be connected to memory chips: a) Two ROM memory chips (each 2k x 16) 11 address lines A0-A10, /ROM (chip select) a) Two RAM memory chips (each 4k x 16) 12 address lines A0-A11, /RAM (chip select), /WRITE The required address space is: ROM 0000H-0FFFH RAM 1000H-2FFFH

Outline the connections for : a) b) c) Simplest linear address decoding, assuming no other devices are on the bus Full address decoding, allowing for system expansion Full address decoding as in (b) but when the above chips are not available and the only chips available are 2k x 8 ROMs and 4k x 8 RAMs

Work out for next week- well go over possible solutions in lecture

Problem Sheet 2 :

General Review

1) Draw and label the block diagram of a small microprocessor system that might be used in a dedicated application. What is the function of each section. 2) Outline the sequence of operations involved in the execution of a typical instruction by a microprocessor such as a 68000. Provide a labelled timing diagram. 3) Explain the differences between (a) operation-code (programme word) fetch cycle, (b) data memory read cycle, and (c) data memory write cycle. 4) What is meant by a wait state and how does its use affect the microprocessor bus control signals and memory cycle times. What devices initiate wait states? 5) Discuss the operation and use of the programme counter and stack pointer and show how they control the sequence of programme execution. 6) Explain the difference between static and dynamic RAM. What are the different types of non-volatile memory? 7) A particular peripheral chip contains four internal address locations which may be written to or read from. Draw a diagram of the bus connections necessary to connect two such chips to a microprocessor using (a) linear addressing (b) fully decoded addressing. 8) Explain the operation of data latches and buffers. How are these used for microprocessor input and output ports? 9) What functions can be performed by counter-timer chips and how might these be used?

Problem Sheet 3 :

General Review

1) Explain the time sequence of actions that occur when a subroutine is called from the main programme with particular reference to role of the stack pointer. 2) How does an external device interrupt the microprocessor programme? What is meant by interrupt priority and how does the microprocessor control which devices can cause interrupts? 3) Explain the difference between synchronous and asynchronous serial data communication and sketch a typical asynchronous data character. How would you obtain this signal from an 8bit parallel data byte? 4) How does a Universal Asynchronous Receiver/Transmitter interface with (a) external systems, and (b) with its host microprocessor? 5) Explain the operation of three types of digital to analogue convertors(DAC): (a) Potential divider network DAC, (b)Binary weighted input DAC, and (c) R-2R ladder DAC. 6) With the aid of pseudo-code or flow charts explain microprocessor programme sequences that use an 8-bit DAC to generate the following analogue signal patterns: (a) square wave whose amplitude is software controlled, (b) a square wave whose period is software controlled (c) a full amplitude sawtooth wave, and (d) a full amplitude triangular wave. 7) How might a microprocessor utilise a timer/counter chip, an input port, and a DAC to provide an analogue sawtooth wave whose period is controlled by an external digital input.

Problem Sheet 4 :

General Review

1) Explain the following types of Analogue to Digital Convertors (ADC) operate: (a) Successive approximation, (b) Flash ADC, and (c) Dual slope ADC. 2) What is meant by aliasing error? What does the Nyquist frequency signify? What is the minimum sampling rate required for signals which contain frequency components up to 20kHz? 3) A microprocessor system has two ADC inputs, one DAC output, and an alarm bell operated by a single output port bit. The two ADC inputs monitor the voltage across an electric motor and the current taken by it. Provide the pseudo-code, or flow chart, for a microprocessor programme that outputs an analogue value proportional to the electric power taken by the motor and also rings an alarm whenever the power taken exceeds a preset maximum value. (hint: power = voltage x current) 4) The motor of question (4) produces many short-term current spikes that cause false alarms. Modify the pseudo- programme/ flow chart to reduce the effects of spikes by averaging over 8 successive current samples. 5) Describe a typical microcontroller and highlight its essential features, identifying their uses. 6) Choose a typical microcontroller application and show how the microcontroller is used. 7) What is Direct Memory Access (DMA) and why is it used? 8) What is Cache memory and what is its purpose? 9) Provide a brief description of different types of microcontroller network: I2C; CANbus; and 1-wire. 10) Outline the essential features of an operating system. What is meant by multi-tasking?